NVIDIA BIGWIG Jen-Hsun Huang has been telling the world plus dog that his glorious graphics processing empire will soon be an equal partner with Intel processors.
In a "today Apple, tomorrow the world" interview reported by CNET, Huang claimed "Apple is an early indicator" of future glories. "The MacBook Pro to the MacBook Air has a GPU," Huang said, and the Air has benefited by having both an Intel CPU and a Nvidia GPU, he claimed.
"Doing the right job with the right tool is more efficient," he said. Although he seemed to forget that under his glorious plan he wants two 'right' tools under the bonnet of the laptop rather than one. He claimed that the Air runs longer and cooler with a GPU.
Nvidia's killer application is OpenCL, which Huang said makes it possible for developers to efficiently tap the vast gigaflops of computing power in the graphics processing unit.
Huang also said his firm's Tegra chip, which is an ARM-based design that integrates an Nvidia GeForce processor, might represent half of Nvidia's business in a few years. The rest will be divided up between the Tesla, Quadro for professionals and the consumer GeForce markets.
He repeated his slur on the business antics of Intel claiming that it is using pricing to prevent Nvidia from selling more of its Ion processors to punters. Huang whinged that Ion would be selling like hotcakes if Intel hadn't been interfering. µ
nVidia overprice their stuff but that wont stop Intel getting another slap for antitrust.
nVidia need to get into cheap lightweight devices to survive and they know it, but given their applesque problems with an inflated sense of self-worth and a rep for cutting corners I wouldn't buy one.
If NV does not manage to completely outperform AMD/ATi in the graphics front, then it is totally out of business. Bad practices or not, they (as anyone else) just can't rely on the Intel front to make a business. They had a good record with their chipsets, but simply when playing with Intel, there is no chance to make a long-standing and stably profitable business there. Intel can leave them out of job at any moment (as their troubles with Ion prove).
What they could do? Here are a few ideas that could be discussed here :
1. if they want a viable chipset division, then they should find ways to innovate more. The integration of an ethernet accelerator and a hardware firewall was a good start that they should pursue more. Why not offer a cipher engine (like Sun in their CPUs), a true hardware RAID on chip and real hardware DSP for audio?
Another good idea would be to pursue chipsets for non-x86 processors (power, sparc, mips, etc)
2. if they can't sell an Ion platform because they need Intel's Atom, then go make a low-power processor. How about using PowerPC architecture that's free? Or even partner with IBM or Freescale to provide a complete solution?
3. enter the licencing world in the ARM model. Why not compete with PowerVR?
He behaves like that crazy old coot running North Korea, looks like a younger version too.
Also has a very distorted view of reality.
I have put him on the "B" Ark list ... he can be shipped off the planet along with the other telephone sanitizers.
i'm not sure what this guy has done for Nvidia, save bad PR, and general corporate shame.
i've been NV all the way since i went from Voodoo2 to TNT2 through to today.
But please, NV, get rid of this joker.
If i had the money, i'd buy NV off this numpty, and even with my zero CEO and management skills, i'm under the impression that i couldn't do any worse.
(i'd probably buy up whats left of Amiga too, and find all the original devs, pump some R&D and release a modern-day amiga500 gaming monster, that you can learn music/GFX/3D/programming on).
And where is OpenCL- enabled nvidia CUDA and/or drivers?
I'm an early CUDA adopter/developer. As far my experience is, I don't want to place all my eggs in one basket, so while CUDA has is positives, it's not the thing I chase for long standing solutions. OpenCL is cool, as it (seems to) be backed by all other major players (minus MicroSoft) - AMD, Apple, (IBM-Cell ?) and Intel's Larrabee.
Imagine my disappointment, as we are cool six (and more) months after the spec. was published, and still no usable implementation (even alpha/beta or RC) is on the market.
So, move along, only PR here...
P.S. They (NVIDIA) didn't even honor my request for closed beta/alpha OpenCL technology preview. I'm just amazed of this lack of vision and corporate wisdom.
nVidia is one of the main contributors to the development of the initial OpenCL spec.... get your facts and opinions straight, people.
I thought Ion was simply a platform/chipset with a nVidia IGP upon which the Intel Atom processor runs? Ol' Jen-Hsun seems not to know what his own products do. There's no excuse as he was whining a couple of weeks ago about being right royally shafted on Atom pricing. Perhaps he meant Atom processors running on Ion based boards. Who knows? Actually, who cares?
At least these airings of his rantings allow the rest of nVidia to actually concentrate once in a while without having to peer into the strategically placed observation mirrors to ensure a visit from the madman isn't imminent.
An aside: OpenCL. I do hope your product has passed Khronos' (no relation) conformance testing if you're name-dropping like that, old son.
http://www.khronos.org/opencl/adopters/
God, he's a lunatic. I said it a year ago that nVidia had wasted their money with PhysX. The way the interest in real time physics was going, there wasn't going to be a place for proprietary GPUs to handle it. There had to be a standard. Yeah, they might've sold a few more cards (quite literally, a few- I'm not sure anybody but hardcore enthusiasts really cared), but as soon as openCL comes along, PhysX is useless. I'm not a hater btw (I have SLI in my laptop).
Nvidea are being smart they know they need OpenCL and not CUDA as OpenCL is backed by Apple and that usually means that it will be the standard as has been shown many times in the past Apple are the trend setter.
And @Drew, OpenCL is a killer app for Intel, AMD/ATi, and Nvidea. Stating it as Nvidea's killer app is not a issue really.
What happened to CUDA and PhysX? No longer killer apps? DAAMIT was first to advertise OpenCL. Now nvidia is just trying to steal the limelights.
"Nvidia's killer application is OpenCL"
Huh?
OpenCL also runs on DAAMIT's chips.
Is "the future" missing chipsets?